The aim of radio network design is to define a set of site locations (i.e., access node locations) and their respective configurations so that the coverage and capacity requirements derived from the dimensioning (i.e., an initial phase of network planning) are met. This has typically been driven by planning tools that rely on radio wave propagation models such as Okumura-Hata and ray tracing models. However, the accuracy of the propagation models can be quite limited, especially in complex urban scenarios. Moreover, each cell can have different propagation characteristics defined by the surrounding environment, and often there are only a handful of propagation models tuned for particular scenarios available. Also, the accuracy of the antenna placement in the planning tool can further contribute to inaccuracies in the predictions. Therefore, there is a need for a more accurate way for performing radio network design capable of overcoming or alleviating at least some of the aforementioned problems.